@exulu/backend 1.57.0 → 1.58.0

package/ee/python/documents/processing/doc_processor.ts

@@ -630,7 +630,7 @@ async function processDocument(
     source: filePath,
   }
 
-  const stripped = filePath.split('.').pop()?.trim();
+  const stripped = filePath.split('.').pop()?.trim().toLowerCase();
   let result: ProcessorOutput;
   switch (stripped) {
     case 'txt':
@@ -1017,7 +1017,7 @@ export async function documentProcessor({
         supportedTypes = ['pdf', 'docx', 'doc', 'txt', 'md', 'jpg', 'jpeg', 'png', 'gif', 'webp'];
         break;
       case "officeparser":
-        supportedTypes = [];
+        supportedTypes = ['docx', 'pptx', 'xlsx', 'odt', 'odp', 'ods', 'pdf', 'rtf', 'csv', 'md', 'html'];
         break;
       case "liteparse":
         supportedTypes = ['pdf', 'doc', 'docx', 'docm', 'odt', 'rtf', 'ppt', 'pptx', 'pptm', 'odp', 'xls', 'xlsx', 'xlsm', 'ods', 'csv', 'tsv'];
@@ -1027,8 +1027,8 @@ export async function documentProcessor({
         break;
     }
 
-    if (!supportedTypes.includes(fileType)) {
-      throw new Error(`[EXULU] Unsupported file type: ${fileType} for Exulu document processor, the ${config?.processor.name} processor only supports the following file types: ${supportedTypes.join(', ')}.`);
+    if (!supportedTypes.includes(fileType.toLowerCase())) {
+      throw new Error(`[EXULU] Unsupported file type: ${fileType.toLowerCase()} for Exulu document processor, the ${config?.processor.name} processor only supports the following file types: ${supportedTypes.join(', ')}.`);
     }
 
     // Process document with VLM validation enabled
@@ -1043,7 +1043,6 @@ export async function documentProcessor({
 
     return content.json;
 
-
   } catch (error) {
     console.error('Error during chunking:', error);
     throw error;

package/ee/python/requirements.txt

@@ -2,3 +2,4 @@ docling
 transformers
 pyinstaller
 docling-hierarchical-pdf
+defusedxml

package/ee/workers.ts

@@ -405,6 +405,9 @@ export const createWorkers = async (
               // not part of the database, so remove it here before
               // we upadte the item in the db.
               delete processorResult.field;
+              // fts is a generated column (tsvector GENERATED ALWAYS AS ... STORED)
+              // and Postgres rejects any explicit update to it.
+              delete processorResult.fts;
 
               // Memory optimization: For large processor results (e.g., documents),
               // extract only the fields we need for the database update to avoid

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@exulu/backend",
   "author": "Qventu Bv.",
-  "version": "1.57.0",
+  "version": "1.58.0",
   "main": "./dist/index.js",
   "private": false,
   "publishConfig": {
@@ -88,8 +88,8 @@
   "dependencies": {
     "@ai-sdk/anthropic": "^3.0.23",
     "@ai-sdk/azure": "^3.0.53",
-    "@ai-sdk/cerebras": "^2.0.29",
-    "@ai-sdk/google-vertex": "^4.0.28",
+    "@ai-sdk/cerebras": "^2.0.51",
+    "@ai-sdk/google-vertex": "^4.0.136",
     "@ai-sdk/openai": "^3.0.18",
     "@ai-sdk/openai-compatible": "^2.0.37",
     "@anthropic-ai/sandbox-runtime": "^0.0.49",